Layout of plots with a unique legend using ggplot - r

Question

I was trying to create a layout with plots sharing the same legend. The legend is on the top of the first plot, however, the next plot has a different scale. How can I solve this?
library(ggplot2)
library(gridExtra)
grid.arrange(
ggplot(mpg, aes(displ, cty)) +
geom_point(aes(shape = "Data")) +
stat_smooth(aes(linetype = "Regression"), method = "lm",
formula = y ~ x, se = FALSE, colour = 1, size = 0.5) +
scale_shape_manual(values = 1) +
labs(shape = "", linetype = "") +
theme_classic() +
theme(panel.border = element_rect(colour = "black", fill=NA, size = 0.5),
aspect.ratio = 1, axis.text = element_text(colour = 1, size = 10),
legend.position = "top")
,
ggplot(mpg, aes(displ, cty)) +
geom_point(shape = 1) +
stat_smooth(method = "lm",
formula = y ~ x, se = FALSE, colour = 1, size = 0.5) +
theme_classic() +
theme(panel.border = element_rect(colour = "black", fill=NA, size = 0.5),
aspect.ratio = 1, axis.text = element_text(colour = 1, size = 10))
)

Answer 1

If the plots also have the same axes labels, facet_wrap may be a good option.
library(ggplot2)
data = rbind(data.frame("id" = 1, mpg), data.frame("id" = 2, mpg))
ggplot(data, aes(displ, cty)) +
geom_point(aes(shape = "Data")) +
stat_smooth(aes(linetype = "Regression"), method = "lm",
formula = y ~ x, se = FALSE, colour = 1, size = 0.5) +
scale_shape_manual(values = 1) +
labs(shape = "", linetype = "") +
theme_classic() +
facet_wrap(~id, ncol = 1 ) +
theme(panel.border = element_rect(colour = "black", fill=NA, size = 0.5),
aspect.ratio = 1, axis.text = element_text(colour = 1, size = 10),
legend.position = "top",
strip.background = element_blank(),
strip.text.x = element_blank()) #these two lines remove the facet strips

Answer 2

grid.arrange doesn't try to align plot panels; it's a generic function meant for all kinds of grid graphics, and in this case since the top plot has a legend it gets shrunk to fit in the available space (by default 1/2 of the page here). For the specific case of ggplots I would use egg::ggarrange,
library(ggplot2)
library(egg)
ggarrange(
ggplot(mpg, aes(displ, cty)) +
geom_point(aes(shape = "Data")) +
stat_smooth(aes(linetype = "Regression"), method = "lm",
formula = y ~ x, se = FALSE, colour = 1, size = 0.5) +
scale_shape_manual(values = 1) +
labs(shape = "", linetype = "") +
theme_classic() +
theme(panel.border = element_rect(colour = "black", fill=NA, size = 0.5),
aspect.ratio = 1, axis.text = element_text(colour = 1, size = 10),
legend.position = "top")
,
ggplot(mpg, aes(displ, cty)) +
geom_point(shape = 1) +
stat_smooth(method = "lm",
formula = y ~ x, se = FALSE, colour = 1, size = 0.5) +
theme_classic() +
theme(panel.border = element_rect(colour = "black", fill=NA, size = 0.5),
aspect.ratio = 1, axis.text = element_text(colour = 1, size = 10))
)

Answer 3

I don't know how to use grid.arrange, but here's a solution using my cowplot package. The idea is to separate the legend out from the plot and then put the three elements into one column. A similar approach would work with grid.arrange, I assume.
library(cowplot)
p1 <- ggplot(mpg, aes(displ, cty)) +
geom_point(aes(shape = "Data")) +
stat_smooth(aes(linetype = "Regression"), method = "lm",
formula = y ~ x, se = FALSE, colour = 1, size = 0.5) +
scale_shape_manual(values = 1) +
labs(shape = "", linetype = "") +
theme_classic() +
theme(panel.border = element_rect(colour = "black", fill=NA, size = 0.5),
aspect.ratio = 1, axis.text = element_text(colour = 1, size = 10),
legend.position = "top")
p2 <- ggplot(mpg, aes(displ, cty)) +
geom_point(shape = 1) +
stat_smooth(method = "lm",
formula = y ~ x, se = FALSE, colour = 1, size = 0.5) +
theme_classic() +
theme(panel.border = element_rect(colour = "black", fill=NA, size = 0.5),
aspect.ratio = 1, axis.text = element_text(colour = 1, size = 10))
legend <- get_legend(p1)
plot_grid(legend, p1 + theme(legend.position = "none"), p2,
ncol=1, rel_heights = c(0.1, 1, 1))